Hen-Wai Tsao

The application of cepstral coefficients and maximum likelihood method in EMG pattern recognition [movements classification]

A new technique for classifying patterns of movement via electromyographic (EMG) signals is presented. Two methods (conventional autoregressive (AR) coefficients and cepstral coefficients) for extracting features from EMG signals and three classification algorithms (Euclidean Distance Measure (EDM), Weighted Distance Measure (WDM), and Maximum Likelihood Method (MLM)) for discriminating signals representative of broad classes of movements are described and compared. These three classifiers are derived from Bayes classifier with some assumptions, the relationship among them is discussed. The conventional MLM is modified to avoid heavy matrix inversion. Six able-bodied subjects with two pairs of surface electrodes located on bilateral sternocleidomastoid and upper trapezius muscles were studied in the experiment. The EMG signals of 20 repetitions of 10 motions were analyzed for each subject. Experimental results showed that mean recognition rate of the cepstral coefficients was at least 5% superior to that of the AR coefficients. The improvement achieved by the cepstral method was statistically significant for all the three classifiers. Reasons for the superiority of cepstral features were investigated from the feature space and frequency domain, respectively. The cepstral coefficients owned better cluster separability in feature space and they emphasized the more informative part in the frequency domain. The discrimination rate of the MLM was the highest among three classifiers. Incorporation of the cepstral features with the MLM could reduce the misclassification rate by 10.6% when compared with the combination of AR coefficients and EDM. Proper choice of five of ten motions could further raise the recognition rate to more than 95%.

A 110-MHz 84-dB CMOS programmable gain amplifier with integrated RSSI function

This paper describes a CMOS programmable gain amplifier (PGA) that maintains a 3-dB bandwidth greater than 110 MHz and can provide an 84-dB gain control range with 1-dB step resolution. The PGA can also be operated in a low-power mode with 3-dB bandwidth greater than 71 MHz. Integrated with this PGA is a CMOS successive logarithmic detecting amplifier with a /spl plusmn/0.7-dB logarithmic accuracy over an 80-dB dynamic range. It achieves -83-dBm sensitivity and consumes 13 mA from a single 3-V supply in the normal power mode. The chip area, including pads, occupies 1.5/spl times/1.5 mm/sup 2/.

The single CCII biquads with high-input impedance

Two new configurations for the design of single second-generation current conveyor (CCII) biquad (SCB) filters with high-input impedance are presented. They can synthesize low-pass, bandpass, high-pass, all-pass, and notch filtering functions with a single CCII connected to five passive one-port RC elements. The active and passive sensitivities are calculated in order to determine the suitable domain for filtering applications. The quality factor and natural frequency of the proposed SCBs are insensitive to the voltage tracking error of the CCII. Experimental results that confirm the theoretical analysis were obtained. The circuits studied show the versatility of the CCIIs. >

A radio-over-fiber network for microcellular system application

This paper investigates the feasibility of a radio-over-fiber network, which employs a single high-power Nd:YAG laser shared among many microcells. External amplitude modulators are employed at both base station (BS) and radio ports (RPs) to form an optical fiber distribution network. The distortions induced by the uplink remodulation are analyzed. The optimal preamplifier gain in the uplink is estimated for given sensitivity and dynamic range (DR). Appropriate modulator linearization and thermal-noise suppression schemes are utilized together to improve the DR. The clipping distortion for the predistortion-linearized modulator is also considered. The result shows that the effects of clipping and higher order distortions have little impact on the estimation of the system performance. A numerical example shows that a 100-mW Nd:YAG laser can provide for a macrocell of 5-km radius up to 16 microcells and 1600 channels when a 20-dB modulator linearization and 10-dB thermal-noise suppression are employed. The experiment setup uses a two-tone test to verify the theoretical calculation. The measurement agrees with the theoretical estimation very well.

Comparison of touch- and laser heat-evoked cortical field potentials in conscious rats

Field potentials and multiunit activities from chronically implanted cortical electrodes were used to study tactile and nociceptive information processing from the tail of the rat. Fourteen stainless steel screws implanted in the skull were used as electrodes to record field potentials in different cortical areas. Electrical, mechanical, and laser pulses were applied to the tail to induce evoked cortical field potentials. Evoked responses were compared before and after sodium pentobarbital anesthesia (50 mg/kg, i.p.). In both electrical- and mechanical-evoked potential (EEP and MEP) studies, two major peaks were found in the conscious animal. The polarity of the late component was modified after pentobarbital anesthesia. In the laser-evoked potential (LEP) study, two distinct negative peaks were found. Both peaks were very sensitive to anesthesia. Following quantitative analysis, our data suggest that the first positive peak of EEP and MEP corresponded to the activation of the Abeta fiber, the second negative peak of MEP and the first peak of LEP corresponded to Adelta fiber activation, while the second peak of LEP corresponded to C fiber activation. The absolute magnitudes of all cortical components were positively related to the intensity of the stimulation. From spatial mapping analysis, a localized concentric source of field potential was observed in the primary somatosensory cortex (SI) only after activation of the Abeta fiber. Larger responsive cortical areas were found in response to Adelta and C fiber activation. In an intracortical recording experiment, both tactile and nociceptive stimulation evoked heightened unit activity changes at latencies corresponding to respective field potentials. We conclude that different cortical areas are involved in the processing of A and C fiber afferent inputs, and barbiturate anesthesia modifies their processing.

Design and Analysis of CMOS Frequency Dividers With Wide Input Locking Ranges

A millimeter-wave (MMW) frequency synthesizer needs a low-power frequency divider (FD) with a wide input locking range to ensure reliability and lower power operation. In this paper, the design and analysis of low-power wide locking range MMW FDs are presented. Proposed are two divide-by-2 (D2) and divide-by-4 (D4) FDs that achieve the widest locking range reported to date by using a dual-mixing technique. Both FDs are fabricated in 90-nm CMOS and are demonstrated to achieve very wide input locking ranges without any tuning mechanism. At an input power of 0 dBm, the D2 FD has a locking range of 51-74 GHz, and that of the D4 FD is 82.5-89 GHz. The power consumption is only 3 mW for both the D2 FD and the D4 FD, from a 0.5 V supply. The proposed D2 and D4 FDs may facilitate incorporation into a product of a MMW phase-locked loop that is smaller, consumes less power, and is more reliable than the conventional approach.

Spectral amplitude-coding optical CDMA system using Mach-Zehnder Interferometers

In this paper, we propose a family of newly constructed codes to suppress the phase-induced intensity noise (PIIN) in spectral amplitude-coding (SAC) optical code division multiple access (OCDMA) systems. These new codes are derived from modified prime codes and their cross-correlation is not larger than one. We also present a novel SAC-OCDMA system employing the new codes together with Mach-Zehnder interferometers to eliminate the multi-user interference (MUI). Compared with the systems employing modified quadratic congruence codes (MQC codes), numerical results verify that our proposed system can more effectively suppress the PIIN and eliminate MUI. Hence, the number of simultaneously users and total transmission rate increases significantly.

Multilevel LINC System Design for Wireless Transmitters

Linear amplifier with nonlinear components (LINC) is a power linearization method which offers both high linearity and high power amplifier (PA) efficiency in wireless transmitters. While LINC increases the power efficiency of PAs, this linearization technique requires an extra power combiner which results in low power efficiency of whole system. To improve this drawback, we propose a multi-level LINC (ML-LINC) method to not only increase power combiner efficiency but also maintain high linearity of wireless transmitters. We also derive the optimal value of each scaling level to maximize the power combiner efficiency. Finally, we demonstrate a four-level scaling ML-LINC as a design example which enhances power combiner efficiency from 44.5% to 80.8% and maintains high linearity to fulfill WCDMA specifications.

A multichannel system for recording and analysis of cortical field potentials in freely moving rats

A system has been developed to record and analyze the cortical electrical activity from 16 different sites in freely moving rats. The hardware includes a 16-channel amplifier system whose high input impedance, low noise, small size, light weight and shielded multistrand connecting cable allow high quality multichannel recording of field potentials. The software developed for this system consists of data acquisition, data analysis and topographic mapping of cortical-evoked potentials as well as electroencephalograms. Cortical field potentials evoked by CO2-laser stimulation were compared between wakeful and pentobarbital-treated conditions. To investigate the background interference produced by sleep spindle, three kinds of reference-free methods (the Wilson, local average and weighted average methods) were utilized to compare the coherence between field potentials obtained from two cerebral hemispheres using monopolar vs. reference-free recordings.

Low-power clock-deskew buffer for high-speed digital circuits

An IC containing four clock deskew buffers using the delay-locked-loop technology has been fabricated in a 0.6 /spl mu/m single poly double metal CMOS process. The core chip area is 0.9/spl times/0.9 mm/sup 2/. The maximum operating frequency is 80 MHz, and the total power dissipation of the four deskew buffers is 59 mW for a 3 V supply voltage. The maximum clock skew after deskewing is less than 300 ps, and the peak-to-peak clock jitter is less than 170 ps. The deskew range is 0.5-3.8 ns.